PyTorchKR Papers

최신 논문

총 5,575편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:1903.03536

Inductive Transfer for Neural Architecture Optimization

The recent advent of automated neural network architecture search led to several methods that outperform state-of-the-art human-designed architectures. However, these approaches are computationally expensive, in extreme cases consuming GPU years. We propose two novel methods which aim to expedite this optimization problem by transferring knowledge acquired from previous tasks to new ones. First, we propose a novel neural architecture selection method which employs this knowledge to identify strong and weak characteristics of neural architectures across datasets. Thus, these characteristics do not need to be rediscovered in every search, a strong weakness of current state-of-the-art searches. Second, we propose a method for learning curve extrapolation to determine if a training process can be terminated early. In contrast to existing work, we propose to learn from learning curves of architectures trained on other datasets to improve the prediction accuracy for novel datasets. On five different image classification benchmarks, we empirically demonstrate that both of our orthogonal contributions independently lead to an acceleration, without any significant loss in accuracy.

arXiv:1903.03462공식 코드 · Papers with Code

On Boosting Semantic Street Scene Segmentation with Weak Supervision

Training convolutional networks for semantic segmentation requires per-pixel ground truth labels, which are very time consuming and hence costly to obtain. Therefore, in this work, we research and develop a hierarchical deep network architecture and the corresponding loss for semantic segmentation that can be trained from weak supervision, such as bounding boxes or image level labels, as well as from strong per-pixel supervision. We demonstrate that the hierarchical structure and the simultaneous training on strong (per-pixel) and weak (bounding boxes) labels, even from separate datasets, constantly increases the performance against per-pixel only training. Moreover, we explore the more challenging case of adding weak image-level labels. We collect street scene images and weak labels from the immense Open Images dataset to generate the OpenScapes dataset, and we use this novel dataset to increase segmentation performance on two established per-pixel labeled datasets, Cityscapes and Vistas. We report performance gains up to +13.2% mIoU on crucial street scene classes, and inference speed of 20 fps on a Titan V GPU for Cityscapes at 512 x 1024 resolution. Our network and OpenScapes dataset are shared with the research community.

arXiv:1903.01998

Statistically-informed deep learning for gravitational wave parameter estimation

We introduce deep learning models to estimate the masses of the binary components of black hole mergers, $(m_1,m_2)$, and three astrophysical properties of the post-merger compact remnant, namely, the final spin, $a_f$, and the frequency and damping time of the ringdown oscillations of the fundamental $\ell=m=2$ bar mode, $(\omega_R, \omega_I)$. Our neural networks combine a modified $\texttt{WaveNet}$ architecture with contrastive learning and normalizing flow. We validate these models against a Gaussian conjugate prior family whose posterior distribution is described by a closed analytical expression. Upon confirming that our models produce statistically consistent results, we used them to estimate the astrophysical parameters $(m_1,m_2, a_f, \omega_R, \omega_I)$ of five binary black holes: $\texttt{GW150914}, \texttt{GW170104}, \texttt{GW170814}, \texttt{GW190521}$ and $\texttt{GW190630}$. We use $\texttt{PyCBC Inference}$ to directly compare traditional Bayesian methodologies for parameter estimation with our deep-learning-based posterior distributions. Our results show that our neural network models predict posterior distributions that encode physical correlations, and that our data-driven median results and 90$\%$ confidence intervals are similar to those produced with gravitational wave Bayesian analyses. This methodology requires a single V100 $\texttt{NVIDIA}$ GPU to produce median values and posterior distributions within two milliseconds for each event. This neural network, and a tutorial for its use, are available at the $\texttt{Data and Learning Hub for Science}$.

arXiv:1903.02440공식 코드 · Papers with Code

SpykeTorch: Efficient Simulation of Convolutional Spiking Neural Networks with at most one Spike per Neuron

Application of deep convolutional spiking neural networks (SNNs) to artificial intelligence (AI) tasks has recently gained a lot of interest since SNNs are hardware-friendly and energy-efficient. Unlike the non-spiking counterparts, most of the existing SNN simulation frameworks are not practically efficient enough for large-scale AI tasks. In this paper, we introduce SpykeTorch, an open-source high-speed simulation framework based on PyTorch. This framework simulates convolutional SNNs with at most one spike per neuron and the rank-order encoding scheme. In terms of learning rules, both spike-timing-dependent plasticity (STDP) and reward-modulated STDP (R-STDP) are implemented, but other rules could be implemented easily. Apart from the aforementioned properties, SpykeTorch is highly generic and capable of reproducing the results of various studies. Computations in the proposed framework are tensor-based and totally done by PyTorch functions, which in turn brings the ability of just-in-time optimization for running on CPUs, GPUs, or Multi-GPU platforms.

arXiv:1903.02428공식 코드 · Papers with Code

Fast Graph Representation Learning with PyTorch Geometric

We introduce PyTorch Geometric, a library for deep learning on irregularly structured input data such as graphs, point clouds and manifolds, built upon PyTorch. In addition to general graph data structures and processing methods, it contains a variety of recently published methods from the domains of relational learning and 3D data processing. PyTorch Geometric achieves high data throughput by leveraging sparse GPU acceleration, by providing dedicated CUDA kernels and by introducing efficient mini-batch handling for input examples of different size. In this work, we present the library in detail and perform a comprehensive comparative study of the implemented methods in homogeneous evaluation scenarios.

arXiv:1903.00757

GraphVite: A High-Performance CPU-GPU Hybrid System for Node Embedding

Learning continuous representations of nodes is attracting growing interest in both academia and industry recently, due to their simplicity and effectiveness in a variety of applications. Most of existing node embedding algorithms and systems are capable of processing networks with hundreds of thousands or a few millions of nodes. However, how to scale them to networks that have tens of millions or even hundreds of millions of nodes remains a challenging problem. In this paper, we propose GraphVite, a high-performance CPU-GPU hybrid system for training node embeddings, by co-optimizing the algorithm and the system. On the CPU end, augmented edge samples are parallelly generated by random walks in an online fashion on the network, and serve as the training data. On the GPU end, a novel parallel negative sampling is proposed to leverage multiple GPUs to train node embeddings simultaneously, without much data transfer and synchronization. Moreover, an efficient collaboration strategy is proposed to further reduce the synchronization cost between CPUs and GPUs. Experiments on multiple real-world networks show that GraphVite is super efficient. It takes only about one minute for a network with 1 million nodes and 5 million edges on a single machine with 4 GPUs, and takes around 20 hours for a network with 66 million nodes and 1.8 billion edges. Compared to the current fastest system, GraphVite is about 50 times faster without any sacrifice on performance.

arXiv:1903.00045

Speeding up Deep Learning with Transient Servers

Distributed training frameworks, like TensorFlow, have been proposed as a means to reduce the training time of deep learning models by using a cluster of GPU servers. While such speedups are often desirable---e.g., for rapidly evaluating new model designs---they often come with significantly higher monetary costs due to sublinear scalability. In this paper, we investigate the feasibility of using training clusters composed of cheaper transient GPU servers to get the benefits of distributed training without the high costs. We conduct the first large-scale empirical analysis, launching more than a thousand GPU servers of various capacities, aimed at understanding the characteristics of transient GPU servers and their impact on distributed training performance. Our study demonstrates the potential of transient servers with a speedup of 7.7X with more than 62.9% monetary savings for some cluster configurations. We also identify a number of important challenges and opportunities for redesigning distributed training frameworks to be transient-aware. For example, the dynamic cost and availability characteristics of transient servers suggest the need for frameworks to dynamically change cluster configurations to best take advantage of current conditions.

arXiv:1707.04610

Cloud-based or On-device: An Empirical Study of Mobile Deep Inference

Modern mobile applications are benefiting significantly from the advancement in deep learning, e.g., implementing real-time image recognition and conversational system. Given a trained deep learning model, applications usually need to perform a series of matrix operations based on the input data, in order to infer possible output values. Because of computational complexity and size constraints, these trained models are often hosted in the cloud. To utilize these cloud-based models, mobile apps will have to send input data over the network. While cloud-based deep learning can provide reasonable response time for mobile apps, it restricts the use case scenarios, e.g. mobile apps need to have network access. With mobile specific deep learning optimizations, it is now possible to employ on-device inference. However, because mobile hardware, such as GPU and memory size, can be very limited when compared to its desktop counterpart, it is important to understand the feasibility of this new on-device deep learning inference architecture. In this paper, we empirically evaluate the inference performance of three Convolutional Neural Networks (CNNs) using a benchmark Android application we developed. Our measurement and analysis suggest that on-device inference can cost up to two orders of magnitude greater response time and energy when compared to cloud-based inference, and that loading model and computing probability are two performance bottlenecks for on-device deep inferences.

arXiv:1902.07316

Deep Modulation Embedding

Deep neural network has recently shown very promising applications in different research directions and attracted the industry attention as well. Although the idea was introduced in the past but just recently the main limitation of using this class of algorithms is solved by enabling parallel computing on GPU hardware. Opening the possibility of hardware prototyping with proven superiority of this class of algorithm, trigger several research directions in communication system too. Among them cognitive radio, modulation recognition, learning based receiver and transceiver are already given very interesting result in simulation and real experimental evaluation implemented on software defined radio. Specifically, modulation recognition is mostly approached as a classification problem which is a supervised learning framework. But it is here addressed as an unsupervised problem with introducing new features for training, a new loss function and investigating the robustness of the pipeline against several mismatch conditions.

arXiv:1902.06155공식 코드 · Papers with Code

Deep Generalized Convolutional Sum-Product Networks

Sum-Product Networks (SPNs) are hierarchical, graphical models that combine benefits of deep learning and probabilistic modeling. SPNs offer unique advantages to applications demanding exact probabilistic inference over high-dimensional, noisy inputs. Yet, compared to convolutional neural nets, they struggle with capturing complex spatial relationships in image data. To alleviate this issue, we introduce Deep Generalized Convolutional Sum-Product Networks (DGC-SPNs), which encode spatial features in a way similar to CNNs, while preserving the validity of the probabilistic SPN model. As opposed to existing SPN-based image representations, DGC-SPNs allow for overlapping convolution patches through a novel parameterization of dilations and strides, resulting in significantly improved feature coverage and feature resolution. DGC-SPNs substantially outperform other SPN architectures across several visual datasets and for both generative and discriminative tasks, including image inpainting and classification. These contributions are reinforced by the first simple, scalable, and GPU-optimized implementation of SPNs, integrated with the widely used Keras/TensorFlow framework. The resulting model is fully probabilistic and versatile, yet efficient and straightforward to apply in practical applications in place of traditional deep nets.

arXiv:1902.06362

Automatic Segmentation of Pulmonary Lobes Using a Progressive Dense V-Network

Reliable and automatic segmentation of lung lobes is important for diagnosis, assessment, and quantification of pulmonary diseases. The existing techniques are prohibitively slow, undesirably rely on prior (airway/vessel) segmentation, and/or require user interactions for optimal results. This work presents a reliable, fast, and fully automated lung lobe segmentation based on a progressive dense V-network (PDV-Net). The proposed method can segment lung lobes in one forward pass of the network, with an average runtime of 2 seconds using 1 Nvidia Titan XP GPU, eliminating the need for any prior atlases, lung segmentation or any subsequent user intervention. We evaluated our model using 84 chest CT scans from the LIDC and 154 pathological cases from the LTRC datasets. Our model achieved a Dice score of $0.939 \pm 0.02$ for the LIDC test set and $0.950 \pm 0.01$ for the LTRC test set, significantly outperforming a 2D U-net model and a 3D dense V-net. We further evaluated our model against 55 cases from the LOLA11 challenge, obtaining an average Dice score of 0.935---a performance level competitive to the best performing team with an average score of 0.938. Our extensive robustness analyses also demonstrate that our model can reliably segment both healthy and pathological lung lobes in CT scans from different vendors, and that our model is robust against configurations of CT scan reconstruction.

arXiv:1902.05093

DeeperLab: Single-Shot Image Parser

We present a single-shot, bottom-up approach for whole image parsing. Whole image parsing, also known as Panoptic Segmentation, generalizes the tasks of semantic segmentation for 'stuff' classes and instance segmentation for 'thing' classes, assigning both semantic and instance labels to every pixel in an image. Recent approaches to whole image parsing typically employ separate standalone modules for the constituent semantic and instance segmentation tasks and require multiple passes of inference. Instead, the proposed DeeperLab image parser performs whole image parsing with a significantly simpler, fully convolutional approach that jointly addresses the semantic and instance segmentation tasks in a single-shot manner, resulting in a streamlined system that better lends itself to fast processing. For quantitative evaluation, we use both the instance-based Panoptic Quality (PQ) metric and the proposed region-based Parsing Covering (PC) metric, which better captures the image parsing quality on 'stuff' classes and larger object instances. We report experimental results on the challenging Mapillary Vistas dataset, in which our single model achieves 31.95% (val) / 31.6% PQ (test) and 55.26% PC (val) with 3 frames per second (fps) on GPU or near real-time speed (22.6 fps on GPU) with reduced accuracy.

arXiv:1902.04610공식 코드 · Papers with Code

Salus: Fine-Grained GPU Sharing Primitives for Deep Learning Applications

GPU computing is becoming increasingly more popular with the proliferation of deep learning (DL) applications. However, unlike traditional resources such as CPU or the network, modern GPUs do not natively support fine-grained sharing primitives. Consequently, implementing common policies such as time sharing and preemption are expensive. Worse, when a DL application cannot completely use a GPU's resources, the GPU cannot be efficiently shared between multiple applications, leading to GPU underutilization. We present Salus to enable two GPU sharing primitives: fast job switching and memory sharing, in order to achieve fine-grained GPU sharing among multiple DL applications. Salus implements an efficient, consolidated execution service that exposes the GPU to different DL applications, and enforces fine-grained sharing by performing iteration scheduling and addressing associated memory management issues. We show that these primitives can then be used to implement flexible sharing policies such as fairness, prioritization, and packing for various use cases. Our integration of Salus with TensorFlow and evaluation on popular DL jobs show that Salus can improve the average completion time of DL training jobs by $3.19\times$, GPU utilization for hyper-parameter tuning by $2.38\times$, and GPU utilization of DL inference applications by $42\times$ over not sharing the GPU and $7\times$ over NVIDIA MPS with small overhead.

arXiv:1902.05020공식 코드 · Papers with Code

Unsupervised 3D End-to-End Medical Image Registration with Volume Tweening Network

3D medical image registration is of great clinical importance. However, supervised learning methods require a large amount of accurately annotated corresponding control points (or morphing), which are very difficult to obtain. Unsupervised learning methods ease the burden of manual annotation by exploiting unlabeled data without supervision. In this paper, we propose a new unsupervised learning method using convolutional neural networks under an end-to-end framework, Volume Tweening Network (VTN), for 3D medical image registration. We propose three innovative technical components: (1) An end-to-end cascading scheme that resolves large displacement; (2) An efficient integration of affine registration network; and (3) An additional invertibility loss that encourages backward consistency. Experiments demonstrate that our algorithm is 880x faster (or 3.3x faster without GPU acceleration) than traditional optimization-based methods and achieves state-of-theart performance in medical image registration.

arXiv:1902.04487공식 코드 · Papers with Code

Extended 2D Consensus Hippocampus Segmentation

Hippocampus segmentation plays a key role in diagnosing various brain disorders such as Alzheimer's disease, epilepsy, multiple sclerosis, cancer, depression and others. Nowadays, segmentation is still mainly performed manually by specialists. Segmentation done by experts is considered to be a gold-standard when evaluating automated methods, buts it is a time consuming and arduos task, requiring specialized personnel. In recent years, efforts have been made to achieve reliable automated segmentation. For years the best performing authomatic methods were multi atlas based with around 80-85% Dice coefficient and very time consuming, but machine learning methods are recently rising with promising time and accuracy performance. A method for volumetric hippocampus segmentation is presented, based on the consensus of tri-planar U-Net inspired fully convolutional networks (FCNNs), with some modifications, including residual connections, VGG weight transfers, batch normalization and a patch extraction technique employing data from neighbor patches. A study on the impact of our modifications to the classical U-Net architecture was performed. Our method achieves cutting edge performance in our dataset, with around 96% volumetric Dice accuracy in our test data. In a public validation dataset, HARP, we achieve 87.48% DICE. GPU execution time is in the order of seconds per volume, and source code is publicly available. Also, masks are shown to be similar to other recent state-of-the-art hippocampus segmentation methods in a third dataset, without manual annotations.

arXiv:1902.04250

Post-Data Augmentation to Improve Deep Pose Estimation of Extreme and Wild Motions

Contributions of recent deep-neural-network (DNN) based techniques have been playing a significant role in human-computer interaction (HCI) and user interface (UI) domains. One of the commonly used DNNs is human pose estimation. This kind of technique is widely used for motion capturing of humans, and to generate or modify virtual avatars. However, in order to gain accuracy and to use such systems, large and precise datasets are required for the machine learning (ML) procedure. This can be especially difficult for extreme/wild motions such as acrobatic movements or motions in specific sports, which are difficult to estimate in typically provided training models. In addition, training may take a long duration, and will require a high-grade GPU for sufficient speed. To address these issues, we propose a method to improve the pose estimation accuracy for extreme/wild motions by using pre-trained models, i.e., without performing the training procedure by yourselves. We assume our method to encourage usage of these DNN techniques for users in application areas that are out of the ML field, and to help users without high-end computers to apply them for personal and end use cases.

arXiv:1902.01878

Disguised-Nets: Image Disguising for Privacy-preserving Outsourced Deep Learning

Deep learning model developers often use cloud GPU resources to experiment with large data and models that need expensive setups. However, this practice raises privacy concerns. Adversaries may be interested in: 1) personally identifiable information or objects encoded in the training images, and 2) the models trained with sensitive data to launch model-based attacks. Learning deep neural networks (DNN) from encrypted data is still impractical due to the large training data and the expensive learning process. A few recent studies have tried to provide efficient, practical solutions to protect data privacy in outsourced deep-learning. However, we find out that they are vulnerable under certain attacks. In this paper, we specifically identify two types of unique attacks on outsourced deep-learning: 1) the visual re-identification attack on the training data, and 2) the class membership attack on the learned models, which can break existing privacy-preserving solutions. We develop an image disguising approach to address these attacks and design a suite of methods to evaluate the levels of attack resilience for a privacy-preserving solution for outsourced deep learning. The experimental results show that our image-disguising mechanisms can provide a high level of protection against the two attacks while still generating high-quality DNN models for image classification.

arXiv:1902.01559

Revisiting a single-stage method for face detection

Although accurate, two-stage face detectors usually require more inference time than single-stage detectors do. This paper proposes a simple yet effective single-stage model for real-time face detection with a prominently high accuracy. We build our single-stage model on the top of the ResNet-101 backbone and analyze some problems with the baseline single-stage detector in order to design several strategies for reducing the false positive rate. The design leverages the context information from the deeper layers in order to increase recall rate while maintaining a low false positive rate. In addition, we reduce the detection time by an improved inference procedure for decoding outputs faster. The inference time of a VGA ($640{\times}480$) image was only approximately 26 ms with a Titan X GPU. The effectiveness of our proposed method was evaluated on several face detection benchmarks (Wider Face, AFW, Pascal Face, and FDDB). The experiments show that our method achieved competitive results on these popular datasets with a faster runtime than the current best two-stage practices.

arXiv:1902.01728

6D Object Pose Estimation without PnP

In this paper, we propose an efficient end-to-end algorithm to tackle the problem of estimating the 6D pose of objects from a single RGB image. Our system trains a fully convolutional network to regress the 3D rotation and the 3D translation in region layer. On this basis, a special layer, Collinear Equation Layer, is added next to region layer to output the 2D projections of the 3D bounding boxs corners. In the back propagation stage, the 6D pose network are adjusted according to the error of the 2D projections. In the detection phase, we directly output the position and pose through the region layer. Besides, we introduce a novel and concise representation of 3D rotation to make the regression more precise and easier. Experiments show that our method outperforms base-line and state of the art methods both at accuracy and efficiency. In the LineMod dataset, our algorithm achieves less than 18 ms/object on a GeForce GTX 1080Ti GPU, while the translational error and rotational error are less than 1.67 cm and 2.5 degree.

arXiv:1902.00842

Real-Time Freespace Segmentation on Autonomous Robots for Detection of Obstacles and Drop-Offs

Mobile robots navigating in indoor and outdoor environments must be able to identify and avoid unsafe terrain. Although a significant amount of work has been done on the detection of standing obstacles (solid obstructions), not much work has been done on the detection of negative obstacles (e.g. dropoffs, ledges, downward stairs). We propose a method of terrain safety segmentation using deep convolutional networks. Our custom semantic segmentation architecture uses a single camera as input and creates a freespace map distinguishing safe terrain and obstacles. We then show how this freespace map can be used for real-time navigation on an indoor robot. The results show that our system generalizes well, is suitable for real-time operation, and runs at around 55 fps on a small indoor robot powered by a low-power embedded GPU.